Techniques of compressing level of detail (lod) data involve defining a cost metric that predicts how much computing resources are necessary to decode and render a mesh at a given lod. The cost metric may be optimized by a selection of a lod reduction process of a plurality of processes at each lod reduction step. For each process of the plurality of processes, the lod is reduced according to that process and the resulting reduced lod is evaluated according to the cost metric. Each such process at that lod reduction step produces a respective lod, which includes a mesh, one or more texture atlases, and/or other attributes. The lod produced by the process having the lowest value of the cost metric at a reduction step is the lod that is input into the next lod reduction step.
|
1. A method, comprising:
obtaining, by processing circuitry of a server computer configured to encode information related to a surface of a three-dimensional object, data representing a first level of detail (lod), the first lod including a first triangular mesh, the first level of detail (lod) being lower than or equal to a reference lod that includes a reference mesh;
performing, by the processing circuitry, a plurality of lod reduction operations on the first lod to produce a plurality of candidate lods and a plurality of recovery bitstreams, each of the plurality of candidate lods including a respective candidate triangular mesh of a plurality of candidate triangular meshes and being lower than the first lod, each of the plurality of recovery bitstreams being associated with a respective candidate lod of the plurality of candidate lods;
performing, by the processing circuitry, a distortion measurement operation on each of the plurality of candidate lods to produce a measure of distortion from the reference lod of a plurality of measures of distortion from the reference lod corresponding to that candidate lod;
generating, by the processing circuitry, a respective value of a lod cost metric associated with each of the plurality of candidate lods, the lod cost metric being based on (i) the measure of distortion of the plurality of measures of distortion corresponding to that candidate lod and (ii) a size of the recovery bitstream associated with that candidate lod; and
performing, by the processing circuitry, a candidate selection operation on the values of the lod cost metric for each of the plurality of candidate lods to produce a second lod, the second lod being lower than the first lod, the candidate selection operation being configured to (i) select, as the second lod, the candidate triangular mesh of the plurality of candidate lods associated with the smallest of the values of the lod cost metric and (ii) select the recovery bitstream of the plurality of recovery bitstreams associated with the second lod, the recovery bitstream enabling a recovery of the first lod from the second lod.
24. An electronic apparatus configured to encode information related to a surface of a three-dimensional object, the electronic apparatus comprising:
memory; and
controlling circuitry coupled to the memory, the controlling circuitry being configured to:
obtain data representing a first triangular mesh for the surface a first level of detail (lod), the first lod including a first triangular mesh, first triangular mesh having the first level of detail (lod) being is lower than or equal to a reference lod that includes of a reference triangular mesh;
perform a plurality of lod reduction operations on the first lod to produce a plurality of candidate lods and a plurality of recovery bitstreams, each of the plurality of candidate lods including a respective candidate triangular mesh of a plurality of candidate triangular meshes and being lower than the first lod mesh, each of the plurality of recovery bitstreams being associated with a respective candidate lod of the plurality of candidate lods;
perform a distortion measurement operation on each of the plurality of candidate lods to produce a measure of distortion from the reference lod of a plurality of measures of distortion from the reference lod corresponding to that candidate lod;
generate a respective value of a lod cost metric associated with each of the plurality of candidate lods, the lod cost metric being based on (i) the measure of distortion of the plurality of measures of distortion corresponding to that candidate lod and (ii) a size of the recovery bitstream associated with that candidate lod; and
perform a candidate selection operation on the values of the lod cost metric for each of the plurality of candidate lods to produce a second lod having a second lod, the second lod being lower less than the first lod, the candidate selection operation being configured to (i) select, as the second lod, the candidate triangular mesh of the plurality of candidate lods associated with the smallest of the values of the lod cost metric and (ii) select the recovery bitstream of the plurality of recovery bitstreams associated with the second lod, the recovery bitstream enabling a recovery of the first lod from the second lod.
23. A computer program product comprising a nontransitory storage medium, the computer program product including code that, when executed by processing circuitry of a user device configured to encode information related to a surface of a three-dimensional object, causes the processing circuitry to perform a method, the method comprising:
obtaining data representing a first level of detail (lod), the first lod including a first triangular mesh, the first level of detail (lod) being lower than or equal to a reference lod that includes a reference mesh;
performing a plurality of lod reduction operations on the first lod to produce a plurality of candidate lods and a plurality of recovery bitstreams, each of the plurality of candidate lods including a respective candidate triangular mesh of a plurality of candidate triangular meshes and being lower than the first lod mesh, each of the plurality of recovery bitstreams being associated with a respective candidate lod of the plurality of candidate lods;
performing a distortion measurement operation on each of the plurality of candidate lods to produce a measure of distortion from the reference lod of a plurality of measures of distortion from the reference lod corresponding to that candidate lod;
generating a respective value of a lod cost metric associated with each of the plurality of candidate lods, the lod cost metric being based on (i) the measure of distortion of the plurality of measures of distortion corresponding to that candidate lod and (ii) a size of the recovery bitstream associated with that candidate lod; and
performing a candidate selection operation on the values of the lod cost metric for each of the plurality of candidate lods to produce a second lod having a second lod, the second lod being lower less than the first lod, the candidate selection operation being configured to (i) select, as the second lod, the candidate triangular mesh of the plurality of candidate lods associated with the smallest of the values of the lod cost metric and (ii) select the recovery bitstream of the plurality of recovery bitstreams associated with the second lod, the recovery bitstream enabling a recovery of the first lod from the second lod.
2. The method as in
4. The method as in
5. The method as in
wherein performing the mesh decimation operation on the first triangular mesh includes performing a batch vertex removal operations on the first triangular mesh.
6. The method as in
wherein performing the mesh decimation operation on the first triangular mesh includes performing a batch of edge collapse operations on the first triangular mesh.
7. The method as in
8. The method as in
9. The method as in
10. The method as in
11. The method as in
generating the set of seam edges from the set of edges of the mesh of the first lod according to an abstraction of the reference lod, the abstraction including a polygonal mesh, the polygonal mesh including a plurality of vertices, a plurality of edges, and a plurality of faces.
12. The method as in
13. The method as in
14. The method as in
generating the abstraction from the reference lod, the abstraction having a measure of distortion from the reference lod that is less than a specified measure of distortion.
15. The method as in
wherein the vertices of an edge of the plurality of edges of the first triangular mesh are on a seam of the set of seams of the abstraction, and
wherein performing the batch of edge-collapse operations includes performing an edge collapse operation on another edge of the plurality of edges of the first triangular mesh.
16. The method as in
wherein the reference lod includes a reference texture atlas, including an embedding of each triangular face of the plurality of triangular faces of the reference lod into a coordinate space of a reference texture image, and
wherein the method further comprises:
performing a re-atlasing operation on the reference texture atlas, the re-atlasing operation providing a conversion of the reference atlas to a texture atlas of the abstraction, to produce, as a texture atlas of the abstraction, an abstracted atlas, the abstracted atlas including a set of patches, each of the set of patches of the abstracted atlas being mapped onto a triangular face of the plurality of triangular faces of the polygonal mesh of the abstraction.
17. The method as in
wherein performing the plurality of lod reduction operations on the first lod includes performing a quantization operation on coordinates of the plurality of vertices of the first triangular mesh and generating a recovery bitstream for the coordinates.
18. The method as in
19. The method as in
wherein the first lod further includes a texture atlas, the texture atlas representing a texture, the texture including a set of patches, each of the set of patches of the texture atlas being mapped onto a triangular face of the plurality of triangular faces of the first triangular mesh, and
wherein performing the plurality of lod reduction operations on the first lod includes performing a quantization operation on coordinates of the set of patches of the texture atlas.
20. The method as in
21. The method as in
22. The method as in
transmitting the first lod, the second lod, and the recovery bitstream to a client computer over a network, the client computer being configured to produce the first lod from the second lod and the recovery bitstream.
|
This description relates to compression of three-dimensional object data.
Some applications such as video games in virtual reality (VR), office applications in augmented reality (AR), and mapping software applications involve representing three-dimensional objects to a user. In one example, an adventure game played by a user in a virtual reality environment may require the generation of virtual trees, rocks, and people. In another example, a mapping application may require the representation of buildings. In some applications, each such object includes a triangular mesh having a plurality of vertices, i.e., points in space that form triangles. Such a triangular mesh involves a plethora of data that may be stored on disk and transmitted to the user. Practical implementations of storing and transmitting the triangular mesh data representing a virtual object include compressing the triangular mesh data. Some applications such as video games in virtual reality (VR), office applications in augmented reality (AR), and mapping software applications involve representing three-dimensional objects to a user. In one example, an adventure game played by a user in a virtual reality environment may require the provide representations of virtual trees, rocks, buildings or people. In some applications, it is beneficial to provide for an object several approximations of varying level of detail. A low level of detail (requiring less data) may be used as a placeholder while the actual object is still loading to improve the experience of the user. In some applications, a lower level is displayed while the object is in the distance, and a higher level is needed, e.g., as the distance between the user and the object decreased. In some applications, data representing these objects include a triangular mesh having a plurality of vertices, i.e., points in space that form triangles, and potentially other attributes such as normals, color or UV coordinates associated with corresponding texture images. A plethora of data that may be stored on disk or transmitted to the user. A progressive encoding of the data is the encoding of a chain of LODs where the encoder provides for a LOD in the chain a data structure that allows the generation of the LOD from a previous LOD in the chain. Thereby, compression can benefit from information in the previous LOD.
In one general aspect, a method can include obtaining, by processing circuitry of a server computer configured to encode information related to a surface of a three-dimensional object, data representing a first level of detail (LOD), the first LOD including a first triangular mesh, the first level of detail (LOD) being lower than or equal to a reference LOD that includes a reference triangular mesh. The method can also include performing, by the processing circuitry, a plurality of LOD reduction operations on the first LOD to produce a plurality of candidate LODs and a plurality of recovery bitstreams, each of the plurality of candidate LODs including a respective candidate triangular mesh of a plurality of candidate triangular meshes and being lower than the first LOD, each of the plurality of recovery bitstreams being associated with a respective candidate LOD of the plurality of candidate LODs. The method can further include performing, by the processing circuitry, a distortion measurement operation on each of the plurality of candidate LODs to produce a measure of distortion from the reference LOD of a plurality of measures of distortion from the reference LOD corresponding to that candidate LOD. The method can further include generating, by the processing circuitry, a respective value of a LOD cost metric associated with each of the plurality of candidate LODs, the LOD cost metric being based on (i) the measure of distortion of the plurality of measures of distortion corresponding to that candidate LOD and (ii) a size of the recovery bitstream associated with that candidate LOD. The method can further include performing, by the processing circuitry, a candidate selection operation on the values of the LOD cost metric for each of the plurality of candidate LODs to produce a second LOD, the second LOD being lower than the first LOD, the candidate selection operation being configured to (i) select, as the second LOD, the candidate LOD of the plurality of candidate LODs associated with the smallest of the values of the LOD cost metric and (ii) select the recovery bitstream of the plurality of recovery bitstreams associated with the second LOD, the recovery bitstream enabling a recovery of the first LOD from the second LOD.
The details of one or more implementations are set forth in the accompanying drawings and the description below. Other features will be apparent from the description and drawings, and from the claims.
A conventional approach to compression of triangular mesh data involves encoding a level of detail (LOD) refinement in a progressive encoder. In this progressive encoder, the refinement is represented by a bitstream produced by the encoder. Such a conventional progressive encoder uses the same algorithm to reduce a LOD at each LOD reduction step. Accordingly, the encoding process begins with a high LOD, including a very fine mesh, from which successively lower LODs, including lower meshes, are all generated using, for example, a specific mesh decimation algorithm to produce a chain of LODs of the encoded progressive mesh.
A technical problem involved in the above-described conventional approach to compressing triangular mesh data is that the above-described conventional approach to compressing triangular mesh data lacks flexibility and may result in a suboptimal decoding experience for a user. For example, a progressive mesh encoder may use the mesh decimation process to define a lower level of detail (LOD) to be used by a progressive mesh decoder, in which a lower LOD corresponds to a less faithful representation of a surface of a three-dimensional object. Nevertheless, there are other LOD reduction processes, e.g., quantization, by which a lower LOD may be defined by an encoder for a decoder. It is unclear which of the different processes may result in a better experience for the user in terms of distortion that the user experiences while the chain of LODs is being transferred and displayed.
A technical solution to the above-described technical problem involves defining a cost metric that predicts how much computing resources are necessary to decode and render a mesh at a given LOD. The cost metric may be optimized by a selection of a LOD reduction process of a plurality of processes at each LOD reduction step. For each process of the plurality of processes, the LOD is reduced according to that process and the resulting reduced LOD is evaluated according to the cost metric. Each such process at that LOD reduction step produces a respective LOD, which includes a mesh, zero or more texture atlases, and/or other attributes. The LOD produced by the process having the lowest value of the cost metric at a reduction step is the LOD that is input into the next LOD reduction step. As an example, three LOD reduction processes include decimation, quantization of vertex coordinates (i.e., “XYZ coordinates”), and quantization of texture coordinates (i.e., “UV coordinates”).
A technical advantage of the above-described technical solution is that a progressive mesh encoder resulting from such an optimization of the agony at each LOD reduction phase will improve the rate-distortion tradeoff and, ultimately, the user's experience in decoding and rendering the LODs representing three-dimensional objects in an application.
As stated above, a LOD as defined herein is a representation of a surface that includes a mesh (e.g., a triangular mesh including connectivity data), one or more texture atlases (i.e., a section of a plane that include texture patches, each of which is mapped onto clusters of faces of the mesh), as well as potentially other attributes (e.g., vertex coordinates, normal, colors). LODs rather than meshes are operated on herein so that cost metrics may be dependent on all data used to describe the object represented by the LOD rather than the mesh only, for instance, the LOD may also be reduced by reducing the quality of the used texture image.
As also stated above, a lower LOD corresponds to a less faithful representation of a surface of a three-dimensional object. Because a representation of a surface includes not only a geometrical mesh but also a texture and attributes, an ordering of the LODs may refer not only to a number of vertices of the mesh, but also a number of pixels of a texture atlas representing a texture of the surface. For example, a first LOD may be higher (or lower) than a second LOD even though their respective meshes have the same number of vertices because the first mesh may have more (or fewer) pixels in their respective texture atlases.
The compression computer 120 is configured to compress data associated with a triangular mesh representing a three-dimensional object. The compression computer 120 includes a network interface 122, one or more processing units 124, and memory 126. The network interface 122 includes, for example, Ethernet adaptors, Token Ring adaptors, and the like, for converting electronic and/or optical signals received from a network to electronic form for use by the point cloud compression computer 120. The set of processing units 124 include one or more processing chips and/or assemblies. The memory 126 includes both volatile memory (e.g., RAM) and non-volatile memory, such as one or more ROMs, disk drives, solid state drives, and the like. The set of processing units 124 and the memory 126 together form control circuitry, which is configured and arranged to carry out various methods and functions as described herein.
In some embodiments, one or more of the components of the compression computer 120 can be, or can include processors (e.g., processing units 124) configured to process instructions stored in the memory 126. Examples of such instructions as depicted in
The LOD manager 130 is configured to obtain, store, and/or transmit first LOD data 132. In some implementations, the LOD manager 130 is configured to receive the first LOD data 132 over a network connection via the network interface 122. In some implementations, the LOD manager 130 is configured to produce the first LOD data 132. In some implementations, the mesh manager 130 is configured to retrieve the first LOD data 132 from a storage medium (e.g., a disk, a flash drive, or the like).
The first LOD data 132 represents a LOD that is input into the LOD reduction manager 140. The LOD itself represents an approximation to a surface of a three-dimensional object. Such an approximation includes a geometric aspect (e.g., mesh connectivity, vertex coordinates), a texture aspect (texture atlas and coordinates), and an attribute aspect (e.g., normal, color). As shown in
The first mesh data 132 represents the triangular mesh that defines the geometry of the approximation to the surface. The mesh includes vertices, edges connecting the vertices, and triangular faces defined by the edges. As shown in
The vertex position data 135 represents the positions of the vertices of the mesh. The positions of the vertices are representations of triplets of real numbers, each triplet being position coordinates of a vertex in space. In some implementations, the position coordinates are quantized so that the real numbers have truncated bit representations. In such an implementation, the vertices are adjusted to lie on a lattice, in which the lattice spacing decreases with the number of bits in the bit representation.
The connectivity data 136 represents the topology and adjacency of the mesh. For example, the connectivity data 136 is used to define a traversal of the mesh. In some implementations, the connectivity data 136 includes triplets of integer indices for each corner of each triangular face of the mesh, each of the triplets including an index identifier of an associated vertex, an index identifier of an opposite corner, and an index identifier of an adjacent corner. The triplets are arranged in an order for traversal of the mesh. Such triplets then provide an ordering of the triangular faces for traversal and an orientation of each triangular face. The positions of the vertices represented by the vertex position data 135 are arranged in an order defined by the index identifiers included in the connectivity data 136. In some implementations, the connectivity data 136 takes the form of a doubly connected edge list, in which an opposing corner is interpreted as a half-edge.
The edge data 137 represents the edges of the mesh as defined by the vertex position data 135. In some implementations, the edges of the mesh are indexed based on the order of the triplets defined by the vertex position data 135.
In some implementations, the texture data 138, if present, represents a texture that may be mapped onto the mesh represented by the first mesh data 134. In some implementations, the texture data 138 includes coordinate pairs within a square or rectangular boundary of a texture atlas. In some implementations, there is more than one texture atlas represented by the texture data 138. In some implementations, the multiple texture atlases share a common UV coordinate system. At each coordinate, there is a scalar quantity representing a brightness or hue of an image. In some implementations, there may be other attributes such as color data for the image; such data may be represented as a triplet or quadruplet of real numbers. In some implementations, the coordinate pairs are quantized, or expressed using a number of bits; that is, the texture atlas includes a grid of coordinates.
The texture data 138 further represents patches of image data such that the data of a patch is mapped to a subset of the triangular faces of the mesh. In some implementations, the texture data 138 also includes seams of the texture atlas, or boundaries of each of the patches. The seams form a subset of the edges represented by the edge data 137.
The LOD reduction manager 140 is configured to perform one of a plurality of LOD reduction processes on the first LOD data 132 to produce the candidate LOD data 150, 150′, and 150″. As shown in
In some implementations, there are one or more instances of the mesh decimation manager 142, each being configured to perform a different mesh decimation operation. In some implementations, the decimation operation may include one of a set of edge collapse operations, such as half-edge, mid-edge or a full edge operations. The full-edge collapse operation allowing to place a merged vertex freely. In some implementations, the full-edge collapse operation is configured such that the merged vertex is placed such that an error function (e.g., a quadric error metric or its variants) is minimized. In some implementations, the above-described decimation operations may include a vertex removal operation
The mesh decimation manager 142 is configured to perform edge collapse operations on the first mesh data 134 to produce mesh data 152 of candidate LOD data 150. In some implementations, the edge collapse operations include a full edge collapse operation and a half-edge collapse operation. The half-edge collapse and the full-edge collapse are discussed in further detail with regard to
In some implementations, the edge collapse operations performed on the first LOD data 132 are configured to preserve a set of seam edges. In some implementations, the seam edges may be provided alongside with the first LOD. In some implementations, the seam edges may be generated from a abstraction of the reference mesh that is provided alongside with the first LOD.
In some implementations, the abstraction may be generated by a separate dedicated process. For instance, in a process that tries to minimize the distortion of the abstraction relative to the provided reference more globally, for instance, for a fixed triangle budget (e.g., a fixed number of triangles). The process may also pay attention to specific properties of the input mesh such as symmetry. In some implementations, the abstraction is provided by an artist to minimize the perceived distortion of a human model by the abstraction.
In some implementations, the abstraction is represented by a set of edges that bound a set of polygonal faces, each being bounded by at least three edges of the set of edges. In some implementations, every polygonal face of the abstraction is bounded by exactly three edges, the abstraction is represented by a triangular mesh.
In some implementations, the set of edges of the triangular mesh of the first LOD include a possibly empty set of boundary edges. The set of seam edges includes the set of boundary edges.
A subdivision is a decomposition of a first set into a second set of subsets such that the intersection of each pair of subsets is empty and such that the union of all subsets equals the first set. Thus, each element of the first set belongs exactly to one subset of the second set. An example of the first set may be a set of all triangles of an input mesh; the subdivision is a set of patches (not necessarily triangular) embedded in a texture image.
A first set and a second set form a one-to-one correspondence if there exists a bijective function from the first set into the second set. That is each element of each set is associated with exactly one element of the other set. Thereby, the first set, or the second set or both may be a set of subsets. Specifically, a set maybe a set of subsets of a subdivision.
The set of seam edges defines a subdivision of the set of triangles of the mesh of the first LOD into a set of connected components. In some implementations, the set of connected components forms a one to one correspondence with the set of polygonal faces of the abstraction.
In some implementations, the set of seam edges is subdivided into chains of edges. The chains are a set of subsets of the set of edges forming a subdivision of the seam edges. The set of subsets form a one to one correspondence with the set of edges of the abstraction.
In some implementation, the set of seam edges may be generated by finding for every vertex of the abstraction a closest vertex on the first LOD. And then finding a disjoint set of chains of edges that connect the found vertices according to the vertex pairs defined by the edges of the abstraction. For instance, one may choose a chain of edges that is closest to the corresponding edges of the abstraction.
In some implementations, a LOD reduction process produces an edge collapse from a predefined set of allowed edge collapses. In some implementations, the LOD reduction process may include an edge collapse into the set of allowed edge collapses if the edge collapse does not move a vertex on a seam.
In some implementations, the LOD reduction process in addition produces an edge collapse into the set of allowed edge collapses if the edge is interior to the chain of edges. An edge is interior to a chain if the edge is not one of the two end edges.
Intuitively, such a seam preserving configuration guides the above-described edge collapse process towards the abstraction since a seam is protected from large distortion by never moving a vertex a way from a seam chain. Specifically, this means that a vertex that belongs to more than one chain of edges does not move during any edge collapse.
In some implementations, the texture data, if present, may be reconfigured such that a chain of edges is mapped onto a straight edge in the two-dimensional set of UV coordinates. The straight edge is for example a corresponding edge of a texture atlas generated for the abstraction. The remaining UV coordinates of each polygonal patch may then be embedded inside a polygon which is the corresponding polygon in the texture atlas of the abstraction such that the embedding of UV coordinates is planar. The texture image information is moved accordingly.
If the first LOD and the second LOD are generated by seam preserving edge collapses, the image of the texture atlas of the first LOD can be shared with the resulting texture atlas of the second LOD.
The XYZ quantization manager 144 is configured to perform an XYZ quantization operation on the vertex position data 135 to produce the candidate data 150′. In some implementations, the XYZ quantization operation includes decrementing the length of a bit string representing each of the coordinate triplets represented by the vertex position data 135 by one bit. In such implementations, a vertex of the mesh is located at a center of a three-dimensional cell of a lattice. The XYZ quantization operation is configured to move the vertex to the center of a three-dimensional cell of a new lattice that has a larger spacing than the previous lattice.
The UV quantization manager 146 is configured to perform a UV quantization operation on the texture data 138, specifically a UV quantization operation on UV coordinate pairs represented by the texture data 138, to produce the candidate data 150″. Similar to the XYZ quantization operation, the UV quantization operation is a reduction in which the length of a bit string representing an approximation to the real numbers of the UV coordinates is decremented by a bit. The UV coordinate of the texture atlas is at the center of a cell of a two-dimensional lattice of UV coordinates. The UV quantization manager then moves the coordinate to the center of a cell of a new lattice that has a larger spacing than the previous lattice.
The candidate LOD data 150, 150′, 150″ represent the output of the above-described LOD reduction operations by the LOD reduction manager 140. Similar to the first LOD data, the candidate LOD data 150 includes respective mesh data 152 and texture data 156, where the mesh data 152 includes vertex position data 153, connectivity data 154, and edge data 155. In addition, the candidate LOD data 150 includes recovery bitstream data 158. The candidate LOD data 150′ and 150″ include similar data as the candidate LOF data 150.
The recovery bitstream data 158 represents a recording of the LOD reduction operation—in this case, the decimation operation—by which the previous LOD may be recovered. The recovery bitstream data 158 is used by a decoder to generate the next higher LOD using the inverse of the LOD reduction operation, in this case the inverse of the mesh decimation operation performed by the mesh decimation manager 142. Further details of the recovery bitstream data 158 is described with regard to
It is noted that the each of the candidate LOD data 150, 150′, 150″ represents an LOD that is lower than the first LOD represented by the first LOD data 132; that is, the each of the candidate data 150, 150′, 150″ has less data and uses fewer computational resources to decode and render than the first LOD data 132. In some implementations, the LOD reduction manager 140 also produces a very low LOD known as an abstraction LOD. The abstraction LOD is used to help define the seams that are to be preserved upon performance of a LOD reduction operation. The abstraction LOD is further discussed with regard to
The cost metric manager 160 is configured to generate cost data 162 for each of the candidate LOD data 150, 150′, 150″. In some implementations, the cost metric manager 160 applies a cost metric (“agony”) to each of the candidate LOD data 150, 150′, 150″. In this implementation, the agony is defined as a product of a measure of distortion corresponding to a candidate LOD (e.g., candidate LOD 150) and the amount of information defining each LOD. In some implementations, the amount of information is a bit rate. In some implementations, the amount of information defining each LOD includes a size of the recovery bitstream associated with that candidate LOD (e.g., recovery bitstream 158). In some implementations, the amount of information defining each LOD is size of the recovery bitstream associated with that candidate LOD. The cost data 162 then includes values of the cost metric generated by the cost metric manager 160.
The selection manager 170 is configured to perform a selection operation on the cost data 162 to produce second LOD data 172. In some implementations, the selection operation includes selecting the smallest value of the agony from the cost data. The second LOD data 172 then represents the LOD associated with that smallest value of the agony.
The second LOD data 172, being one of the candidate LOD data 150, 150′, 150″, includes respective mesh data 174, texture data 178, and a recovery bitstream 179, where the mesh data 174 includes vertex position data 175, connectivity data 176, and edge data 177. The second LOD data 172 represents output of a progressive encoding operation according to the improved techniques disclosed herein. Moreover, the second LOD data 172 may be used as input into another such progressive encoding operation as described herein. These progressive encoding operations may be repeated until a lowest LOD is achieved.
At 202, the LOD manager 130 obtains data (e.g., first LOD data 132) representing a first level of detail (LOD), the first LOD including a first triangular mesh (e.g., mesh data 134), the first level of detail (LOD) being lower than or equal to a reference LOD that includes a reference triangular mesh.
At 204, the LOD reduction manager 140 performs a plurality of LOD reduction operations on the first LOD to produce a plurality of candidate LODs (e.g., candidate LOD data 150, 150′, 150″) and a plurality of recovery bitstreams (e.g., recovery bitstream 158 and recovery bitstreams associated with LOD data 150 and 150″), each of the plurality of candidate LODs including a respective candidate triangular mesh (e.g. mesh data 152 and mesh data associated with LOD data 150 and 150″) of a plurality of candidate triangular meshes and being lower than the first LOD, each of the plurality of recovery bitstreams being associated with a respective candidate LOD of the plurality of candidate LODs.
At 206, the cost metric manager 160 performs a distortion measurement operation on each of the plurality of candidate LODs to produce a measure of distortion from the reference LOD of a plurality of measures of distortion from the reference LOD corresponding to that candidate LOD.
At 208, the cost metric manager 160 generates a respective value of a LOD cost metric associated with each of the plurality of candidate LODs, the LOD cost metric being based on (i) the measure of distortion of the plurality of measures of distortion corresponding to that candidate LOD and (ii) a size of the recovery bitstream associated with that candidate LOD.
At 210, the selection manager 170 performs a candidate selection operation on the values of the LOD cost metric for each of the plurality of candidate LODs to produce a second LOD (e.g., second LOD data 172), the second LOD being lower than the first LOD, the candidate selection operation being configured to (i) select, as the second LOD, the candidate triangular mesh of the plurality of candidate LODs associated with the smallest of the values of the LOD cost metric and (ii) select the recovery bitstream of the plurality of recovery bitstreams associated with the second LOD, the recovery bitstream enabling a recovery of the first LOD from the second LOD.
As illustrated in
In some implementations, to generate the abstraction mesh 312, the LOD reduction manager 140 performs a sequence of mesh decimation operations on the reference mesh 306 until the measure of distortion between the abstraction mesh 312 and the reference mesh 306 is greater than a specified distortion value. In generating the abstraction mesh 312, the LOD reduction manager 140 does not use any information regarding the reference texture atlas 304. Further details about the abstraction mesh 312 and the abstraction texture atlas are disclosed with regard to
Some input textured meshes are generated using automated algorithms. Such algorithms often create a complex texture atlas with fragmented and irregular texture seams. An extreme case occurs when all triangle edges correspond to seams. In other cases, many texture mapping methods are not aware of a LOD and are unfit for effective progressive compression. Such a scenario is illustrated in
To effect the preservation of the seams in a re-atlasing operation, all triangular faces of the input mesh 652 such that each cluster corresponds to one triangular face of the abstraction mesh 662. A texture patch is then formed for each cluster by generating a planar parameterization onto the corresponding abstraction triangle. Depending on the complexity of the abstraction mesh 662, the new (abstraction) texture atlas 604 may still include a large number of seams which constrain decimation operations and are costly to single-rate encode as each seam uses duplicated UV coordinates. Accordingly, the triangular texture patches are packed by attaching the triangles as much as possible so as to exchange a low number of texture seams for limited texture distortion. All texture seams which are attached are first saved as “virtual seams,” which are preserved during the mesh decimation processes. After performing the re-atlasing operation, both the input (reference) mesh 652 and the abstraction mesh 654 are assigned new UV coordinates in the newly generated (abstraction) texture atlas 604, from which the textured abstraction atlas 664 is produced.
Returning to
The progressive encoding operation 318 is configured to produce a stream of refinement via quantization and LOD generation using a fine-to-coarse simplification approach. As an initial step, the progressive encoding operation 318 includes an initial quantization step in which a global, uniform quantization operation is performed for XYZ (vertex) coordinates and UV (texture) coordinates by converting the continuous, real coordinates into discrete, integer coordinates. The XYZ coordinates use Qg bits and the UV coordinates use Qt bits; the initial values of Qg and Qt are large enough to yield a negligible distortion.
The progressive encoding operation 318 is configured to perform a plurality of LOD reduction operations on meshes of LODs to produce meshes of candidate LODs. As shown in
Each of the geometry-based LOD reduction operations, the mesh decimation operation 324, the XYZ quantization step 326, and the UV quantization operations 328, introduces an increase in the measure of distortion ΔD while reversing a LOD reduction operation during a decoding uses a bitrate ΔR. In some implementations, the measure of distortion ΔD is a perceptual distortion metric (i.e., distortions from multiple points of view). In some implementations, the measure of distortion ΔD is an absolute value of a sum over displacements of one of the XYZ and UV coordinates. In some implementations, the measure of distortion ΔD is an absolute value of a sum over displacements of the XYZ and UV coordinates.
In some implementations, the cost metric for each LOD reduction operations for a given LOD is equal to the product of the measure of distortion and the bitrate, ΔDΔR for that LOD reduction operation at that LOD. This particular cost metric is referred herein as the agony. There are other cost metrics depending on the measure of distortion and bitrate. In some implementations, the cost metric is proportional to the product of a power of the measure of distortion and another power of the bitrate.
One may plot the measure of distortion ΔD against the bitrate ΔR to produce a rate-distortion (R-D) curve. The R-D curve usually reflects a tradeoff between the measure of distortion and the bitrate. The R-D curve is represented by stairsteps over the tradeoff curve, such that the measure of distortion ΔD is constant between two LODs.
The cost metric manager 170 writes the values of the measure of distortion ΔD and the bitrate ΔR to the memory 126 (e.g., cost data 162). In some implementations, the cost metric manager 170 writes the agony, or more generally, the values of the cost metric, to the memory 126. The compression computer 120 (e.g., the selection manager 170) then selects, via a candidate selection operation 360, the smallest value of the cost metric from the cost data 162, and the candidate LOD data (e.g., candidate LOD data 334, 336, and 338) associated with that smallest value of the cost metric to produce the reduced LOD 362. It is that candidate LOD data 334, 336, and 338, including the associated mesh data and texture data, that was output from the current LOD reduction operation and may be input into a next LOD reduction operation. In some implementations, the above-described process may be repeated until a specified maximum distortion has been achieved. In some implementations, the specified maximum distortion is the same used to derive the abstraction LOD. It should be noted that the reduced LOD 362 achieved by the above-described process is not necessarily the abstraction LOD as the abstraction LOD is derived by a sequence of mesh decimation operations.
As shown in
The constraint to maintain the input mesh connectivity for the highest LOD leads to a common batched mesh decimation paradigm to be applied. In some implementations, the mesh decimation operations are performed using batches of independent edge collapse operators sorted in a priority queue.
During decoding, a canonical ordering of the vertices of a mesh is used to locate the vertex split operators which reverse (i.e., invert) the edge collapse operators used in the mesh decimation operations, within the current batch. Such a canonical ordering differs from the ordering of edge collapse operators applied during mesh decimation operations, which depends on a priority queue. Accordingly, in some implementations, the collapsed edges produced by the edge collapse operators for an independent set. That is, the collapsed edges are sufficiently separated from each other such that each edge can be collapsed without interfering with the other edges of the plurality of edges (e.g., edge data 155). In such an implementation, an edge cannot be collapsed if one of the vertices of its patch is the result of a previous edge collapse.
The quadric error metric extended to five dimensions to account for XYZ and UV coordinates as described above preserves both input mesh geometry and re-atlased (abstraction) texture coordinates, and also handles texture seams. The computation of this error metric produces, for each edge, an optimal vertex and texture atlas attributes of the resulting vertex upon collapse which minimizes the error defined by the metric. In each batch, the edge collapse operators are organized in a priority queue, where the priority indicates the error introduced by the edge collapse operation. The next mesh decimation operation selected is that which produces the smallest value of the error metric. The mesh decimation operations are stopped when the value of the error metric is greater than a specified percentile (e.g., 30%) of the errors computed for the plurality of edges represented by the edge data of the initial LOD data of the batch.
For the mesh decimation operations, a degree of freedom is a type of collapse (i.e., split) operation: a half-edge collapse operation and a full-edge collapse operation. In the full-edge collapse operation, the location of the merged vertex is determined by minimizing the local error using the above-described error metric, at the cost of encoding two residual vectors between the merged vertex and its two ancestors. In the half-edge collapse operation, the location of the merged vertex is restricted to one of its two ancestors. While the half-edge collapse operations introduce a larger local error than that introduced for the full-edge collapse operation, only a single residual vector is encoded to reverse the operation during decoding.
In some implementations, a mesh decimation operation includes a hybrid of half-edge and full-edge collapse operations. In some implementations, the hybrid includes performing a first batch of half-edge collapse operations and then performing a second batch of full-edge collapse operations after performing the first batch of half-edge collapse operations. Such a hybrid improves the rate-distortion tradeoff because a dense (initial) mesh can achieve lower errors using half-edge collapse operations while coarse, later meshes should use full-edge collapse operations to achieve a lower error. The LOD at which the hybrid switches from half-edge collapse operation to full-edge collapse operation is chosen to minimize the agony.
As indicated in
Returning to
After each batch of mesh decimation operations, the vertices of the mesh are indexed (ordered) according to a canonical spanning tree traversal, and the vertices to split are identified by a sequence of binary symbols. In some implementations, the sequence is shortened by omitting the vertices that cannot be split due to the constraints of having an independent set of vertex split operations. For each vertex split operation, the geometry data (e.g., the mesh data), connectivity data, and texture mapping data are encoded after prediction.
The connectivity data are encoded first to help predict the geometry represented by the geometry data. For example, when performing a vertex splitting operation on a vertex to produce two vertices, recovering the connectivity data includes producing a connection between the two vertices and their neighbors. Specifically, manifold and non-manifold cases are encoded separately because the surface represented by the LOD data remains locally 2-manifold. Because an edge collapse operation can transform a non-manifold connectivity into a manifold connectivity and vice-versa, a binary symbol is used to indicate a topological change. For non-manifold connectivity cases, specified binary symbols are used to indicate which vertices result from an edge collapse operation. For manifold connectivity cases, an umbrella of a split vertex is split into a first part and a second part. The first part is formed by vertices that are connected to only one vertex of an un-collapsed edge, either the first or second, which forms two sets. The second part is formed by the two vertices that delineate the boundary of these two sets which are connected to both edge vertices forming two new triangles. During decoding, these two vertices are identified and its connected vertex assigned to each set. For encoding the boundary vertices of the umbrella, the two boundary vertices that are furthest apart (i.e., the diameter of the umbrella) are used as a predictor for deriving the direction of the split edge. Such a prediction is discussed in further detail with regard to
The locations of the two vertices v1 and v2 after a vertex split operation uses two residual vectors to reverse a full-edge collapse, and one residual vector for reversing a half-edge collapse. The decoded connectivity may be leveraged to perform two independent and sequential barycentric predictions for v1 and then v2 (for a full-edge collapse operation) using the vertices of their respective patch. Once v1 is decoded, then its location is used for a barycentric prediction of v2.
It is noted that the increase in distortion introduced by a coarse quantization is smaller for a coarse mesh than a dense mesh. Along these lines, an adaptive quantization operation includes decreasing the precision of vertex or texture (XYZ or UV) coordinates as the LOD reductions progress. In some implementations, a quantization operation decrements by one the number of quantization bits for a coordinate type (e.g., XYZ or UV). Such a quantization operation is performed when its agony is less than that produced by other types of operations (e.g., mesh decimation). Reversing this operation in decoding involves relocating each vertex to the center of a smaller cell in a denser lattice. Predicting the new locations may be accomplished by generating a distance to a centroid of neighboring vertices of a cell of a parent lattice. The adaptive quantization operation helps improve the compression rate by using single-rate compression with lower quantization bits, shifting the R-D curve to the left.
The UV texture atlas coordinates may be predicted similarly to a parallelogram prediction, which uses the XYZ and UV coordinates of the triangle vertices to predict new texture coordinates.
As shown in
The geometry and texture multiplexing operation is configured to multiplex the progressive geometry and texture data from coarse to dense, as better performance is obtained than by simple proceeding from coarse to dense. This is made possible because the geometry and texture data are independent. Specifically, starting from the LOD that pairs the coarsest mesh with the coarsest texture, a refinement of either the geometry or the texture is selected based on the smallest agony.
Returning to
A geometry refinement may be either a batch of XYZ quantization refinements, a batch of UV quantization refinements, or a batch of mesh refinement operations. In some implementations, the batch or refinement operations may be organized as follows:
Results of the above-described cost-driven progressive encoding framework are discussed with regard to
The above graphs in
As shown in
Computing device 1200 includes a processor 1202, memory 1204, a storage device 1206, a high-speed interface 1208 connecting to memory 1204 and high-speed expansion ports 1210, and a low speed interface 1212 connecting to low speed bus 1214 and storage device 1206. Each of the components 1202, 1204, 1206, 1208, 1210, and 1212, are interconnected using various busses, and may be mounted on a common motherboard or in other manners as appropriate. The processor 1202 can process instructions for execution within the computing device 1200, including instructions stored in the memory 1204 or on the storage device 1206 to display graphical information for a GUI on an external input/output device, such as display 1216 coupled to high speed interface 1208. In other implementations, multiple processors and/or multiple buses may be used, as appropriate, along with multiple memories and types of memory. Also, multiple computing devices 1200 may be connected, with each device providing portions of the necessary operations (e.g., as a server bank, a group of blade servers, or a multi-processor system).
The memory 1204 stores information within the computing device 1200. In one implementation, the memory 1204 is a volatile memory unit or units. In another implementation, the memory 1204 is a non-volatile memory unit or units. The memory 1204 may also be another form of computer-readable medium, such as a magnetic or optical disk.
The storage device 1206 is capable of providing mass storage for the computing device 1200. In one implementation, the storage device 1206 may be or contain a computer-readable medium, such as a floppy disk device, a hard disk device, an optical disk device, or a tape device, a flash memory or other similar solid state memory device, or an array of devices, including devices in a storage area network or other configurations. A computer program product can be tangibly embodied in an information carrier. The computer program product may also contain instructions that, when executed, perform one or more methods, such as those described above. The information carrier is a computer- or machine-readable medium, such as the memory 1204, the storage device 1206, or memory on processor 1202.
The high speed controller 1208 manages bandwidth-intensive operations for the computing device 600, while the low speed controller 1212 manages lower bandwidth-intensive operations. Such allocation of functions is exemplary only. In one implementation, the high-speed controller 1208 is coupled to memory 1204, display 1216 (e.g., through a graphics processor or accelerator), and to high-speed expansion ports 1210, which may accept various expansion cards (not shown). In the implementation, low-speed controller 1212 is coupled to storage device 1206 and low-speed expansion port 1214. The low-speed expansion port, which may include various communication ports (e.g., USB, Bluetooth, Ethernet, wireless Ethernet) may be coupled to one or more input/output devices, such as a keyboard, a pointing device, a scanner, or a networking device such as a switch or router, e.g., through a network adapter.
The computing device 1200 may be implemented in a number of different forms, as shown in the figure. For example, it may be implemented as a standard server 1220, or multiple times in a group of such servers. It may also be implemented as part of a rack server system 1224. In addition, it may be implemented in a personal computer such as a laptop computer 1222. Alternatively, components from computing device 1200 may be combined with other components in a mobile device (not shown), such as device 1250. Each of such devices may contain one or more of computing device 1200, 1250, and an entire system may be made up of multiple computing devices 1200, 1250 communicating with each other.
Computing device 1250 includes a processor 1252, memory 1264, an input/output device such as a display 1254, a communication interface 1266, and a transceiver 1268, among other components. The device 1250 may also be provided with a storage device, such as a microdrive or other device, to provide additional storage. Each of the components 1250, 1252, 1264, 1254, 1266, and 1268, are interconnected using various buses, and several of the components may be mounted on a common motherboard or in other manners as appropriate.
The processor 1252 can execute instructions within the computing device 1250, including instructions stored in the memory 1264. The processor may be implemented as a chipset of chips that include separate and multiple analog and digital processors. The processor may provide, for example, for coordination of the other components of the device 1250, such as control of user interfaces, applications run by device 1250, and wireless communication by device 1250.
Processor 1252 may communicate with a user through control interface 1258 and display interface 1256 coupled to a display 1254. The display 1254 may be, for example, a TFT LCD (Thin-Film-Transistor Liquid Crystal Display) or an OLED (Organic Light Emitting Diode) display, or other appropriate display technology. The display interface 1256 may comprise appropriate circuitry for driving the display 254 to present graphical and other information to a user. The control interface 1258 may receive commands from a user and convert them for submission to the processor 1252. In addition, an external interface 1262 may be provided in communication with processor 1252, so as to enable near area communication of device 1250 with other devices. External interface 1262 may provide, for example, for wired communication in some implementations, or for wireless communication in other implementations, and multiple interfaces may also be used.
The memory 1264 stores information within the computing device 1250. The memory 1264 can be implemented as one or more of a computer-readable medium or media, a volatile memory unit or units, or a non-volatile memory unit or units. Expansion memory 1274 may also be provided and connected to device 1250 through expansion interface 1272, which may include, for example, a SIMM (Single In Line Memory Module) card interface. Such expansion memory 1274 may provide extra storage space for device 1250, or may also store applications or other information for device 1250. Specifically, expansion memory 1274 may include instructions to carry out or supplement the processes described above, and may include secure information also. Thus, for example, expansion memory 1274 may be provided as a security module for device 1250, and may be programmed with instructions that permit secure use of device 1250. In addition, secure applications may be provided via the SIMM cards, along with additional information, such as placing identifying information on the SIMM card in a non-hackable manner.
The memory may include, for example, flash memory and/or NVRAM memory, as discussed below. In one implementation, a computer program product is tangibly embodied in an information carrier. The computer program product contains instructions that, when executed, perform one or more methods, such as those described above. The information carrier is a computer- or machine-readable medium, such as the memory 1264, expansion memory 1274, or memory on processor 1252, that may be received, for example, over transceiver 1268 or external interface 1262.
Device 1250 may communicate wirelessly through communication interface 1266, which may include digital signal processing circuitry where necessary. Communication interface 1266 may provide for communications under various modes or protocols, such as GSM voice calls, SMS, EMS, or MMS messaging, CDMA, TDMA, PDC, WCDMA, CDMA2000, or GPRS, among others. Such communication may occur, for example, through radio-frequency transceiver 1268. In addition, short-range communication may occur, such as using a Bluetooth, WiFi, or other such transceiver (not shown). In addition, GPS (Global Positioning System) receiver module 1270 may provide additional navigation- and location-related wireless data to device 1250, which may be used as appropriate by applications running on device 1250.
Device 1250 may also communicate audibly using audio codec 1260, which may receive spoken information from a user and convert it to usable digital information. Audio codec 1260 may likewise generate audible sound for a user, such as through a speaker, e.g., in a handset of device 1250. Such sound may include sound from voice telephone calls, may include recorded sound (e.g., voice messages, music files, etc.) and may also include sound generated by applications operating on device 1250.
The computing device 1250 may be implemented in a number of different forms, as shown in the figure. For example, it may be implemented as a cellular telephone 1280. It may also be implemented as part of a smart phone 1282, personal digital assistant, or other similar mobile device.
Various implementations of the systems and techniques described here can be realized in digital electronic circuitry, integrated circuitry, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and/or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Various implementations of the systems and techniques described here can be realized in digital electronic circuitry, integrated circuitry, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and/or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.
These computer programs (also known as programs, software, software applications or code) include machine instructions for a programmable processor, and can be implemented in a high-level procedural and/or object-oriented programming language, and/or in assembly/machine language. As used herein, the terms “machine-readable medium” “computer-readable medium” refers to any computer program product, apparatus and/or device (e.g., magnetic discs, optical disks, memory, Programmable Logic Devices (PLDs)) used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term “machine-readable signal” refers to any signal used to provide machine instructions and/or data to a programmable processor.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a client computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (“LAN”), a wide area network (“WAN”), and the Internet.
The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
Returning to
The components (e.g., modules, processing units 124) of the compression computer 120 can be configured to operate based on one or more platforms (e.g., one or more similar or different platforms) that can include one or more types of hardware, software, firmware, operating systems, runtime libraries, and/or so forth. In some implementations, the components of the compression computer 120 can be configured to operate within a cluster of devices (e.g., a server farm). In such an implementation, the functionality and processing of the components of the compression computer 120 can be distributed to several devices of the cluster of devices.
The components of the compression computer 120 can be, or can include, any type of hardware and/or software configured to process attributes. In some implementations, one or more portions of the components shown in the components of the compression computer 120 in
Although not shown, in some implementations, the components of the compression computer 120 (or portions thereof) can be configured to operate within, for example, a data center (e.g., a cloud computing environment), a computer system, one or more server/host devices, and/or so forth. In some implementations, the components of the compression computer 120 (or portions thereof) can be configured to operate within a network. Thus, the components of the compression computer 120 (or portions thereof) can be configured to function within various types of network environments that can include one or more devices and/or one or more server devices. For example, a network can be, or can include, a local area network (LAN), a wide area network (WAN), and/or so forth. The network can be, or can include, a wireless network and/or wireless network implemented using, for example, gateway devices, bridges, switches, and/or so forth. The network can include one or more segments and/or can have portions based on various protocols such as Internet Protocol (IP) and/or a proprietary protocol. The network can include at least a portion of the Internet.
In some embodiments, one or more of the components of the compression computer 120 can be, or can include, processors configured to process instructions stored in a memory. For example, the LOD manager 130 (and/or a portion thereof), the LOD reduction manager 140 (and/or a portion thereof), the cost metric manager 160 (and/or a portion thereof), and the selection manager 170 (and/or a portion thereof) can be a combination of a processor and a memory configured to execute instructions related to a process to implement one or more functions.
A number of embodiments have been described. Nevertheless, it will be understood that various modifications may be made without departing from the spirit and scope of the specification.
For example, in some implementations, the value of the LOD cost metric associated with each of the plurality of candidate LODs includes a difference between a measure of distortion of the first LOD and the measure of distortion corresponding to that candidate LOD divided by the size of the recovery bitstream associated with that candidate LOD.
In some implementations, the value of the LOD cost metric associated with each of the plurality of candidate LODs includes a product of a difference between a measure of distortion of the first LOD and the measure of distortion corresponding to that candidate LOD and the size of the recovery bitstream associated with that candidate LOD.
In some implementations, the reference LOD is the first LOD.
In some implementations, performing the plurality of LOD reduction operations on the first LOD includes performing a mesh decimation operation on the first triangular mesh to produce a candidate LOD and obtaining the recovery bitstream associated with the candidate LOD.
In some implementations, the first triangular mesh further includes a plurality of vertices and a plurality of edges, each of the plurality of edges including a respective pair of vertices of the plurality of vertices of the triangular mesh, and performing the mesh decimation operation on the first triangular mesh includes performing a batch vertex removal operations on the first triangular mesh.
In some implementations, the first triangular mesh further includes a plurality of vertices and a plurality of edges, each of the plurality of edges including a respective pair of vertices of the plurality of vertices of the triangular mesh, and performing the mesh decimation operation on the first triangular mesh includes performing a batch of edge collapse operations on the first triangular mesh.
In some implementations, the batch of edge collapse operations is a batch of half-edge collapses.
In some implementations, the batch of edge collapse operations is a batch of full-edge collapses.
In some implementations, the batch of edge collapse operations is a batch of mid-edge collapses.
In some implementations, the vertices of an edge of the plurality of edges of the first triangular mesh are on a seam of a set of seam edges, and wherein performing the batch of edge-collapse operations includes performing an edge collapse operation on another edge of the plurality of edges of the first triangular mesh.
In some implementations, the method further comprises generating the set of seam edges from the set of edges of the mesh of the first LOD according to an abstraction of the reference LOD, the abstraction including a polygonal mesh, the polygonal mesh including a plurality of vertices and a plurality of edges.
In some implementations, the set of seam edges is selected such that it is a subdivision of edge chains, where the subdivision forms a one to one correspondence to the set of edges of the abstraction.
In some implementations, the set of seam edges is selected such that it subdivides the set of triangles of the mesh of the first LOD into a set of connected components that forms a one to one correspondence to the set of polygonal faces of the abstraction.
In some implementations, the method further comprises generating the abstraction LOD from the reference LOD, the abstraction LOD having a measure of distortion from the reference LOD that is less than a specified measure of distortion.
In some implementations, the first LOD further includes a first texture atlas, the first texture atlas including a set of seams defining boundaries of a set of patches, each of the set of patches including an image configured to be superimposed onto a set of mesh triangles of the first triangular mesh, the vertices of an edge of the plurality of edges of the first triangular mesh are on a seam of the set of seams of the abstraction, and performing the batch of edge-collapse operations includes performing an edge collapse operation on another edge of the plurality of edges of the first triangular mesh.
In some implementations, the abstraction triangular mesh and the reference triangular mesh each include a respective plurality of triangular faces, the reference LOD includes a reference texture atlas, including an embedding of each triangle of the plurality of triangles of the reference LOD into the UV coordinate space of a reference texture image, and the method further comprises performing a re-atlasing operation on the reference texture atlas, providing a conversion of the reference atlas to a texture atlas of the abstraction. for each cluster of the set of clusters, to produce, as a texture atlas of the abstraction LOD, an abstracted atlas, the abstracted atlas including a set of patches, each of the set of patches of the abstracted atlas being mapped onto a triangular face of the plurality of triangular faces of the abstraction triangular mesh.
In some implementations, the first triangular mesh includes a plurality of vertices and performing the plurality of LOD reduction operations on the first LOD includes performing a quantization operation on coordinates of the plurality of vertices of the first triangular mesh and generating a recovery bitstream for the coordinates.
In some implementations, performing the quantization operation includes decrementing a number of bits used in representing the coordinates of the plurality of vertices In some implementations,
In some implementations, the first triangular mesh further includes a plurality of triangular faces, the first LOD further includes a texture atlas, the texture atlas representing a texture, the texture including a set of patches, each of the set of patches of the texture atlas being mapped onto a triangular face of the plurality of triangular faces of the first triangular mesh, and performing the plurality of LOD reduction operations on the first LOD includes performing a quantization operation on coordinates of the set of patches of the texture atlas.
In some implementations, performing the quantization operation includes decrementing a number of bits used in a representation of the coordinates of the set of patches of the texture atlas and generating a recovery bitstream for the coordinates.
In some implementations, the triangular mesh of the second LOD has fewer vertices than the triangular mesh of the first LOD.
In some implementations, the method further comprises transmitting the first LOD, the second LOD, and the recovery bitstream to a client computer over a network, the client computer being configured to produce the first LOD from the second LOD and the recovery bitstream.
It will also be understood that when an element is referred to as being on, connected to, electrically connected to, coupled to, or electrically coupled to another element, it may be directly on, connected or coupled to the other element, or one or more intervening elements may be present. In contrast, when an element is referred to as being directly on, directly connected to or directly coupled to another element, there are no intervening elements present. Although the terms directly on, directly connected to, or directly coupled to may not be used throughout the detailed description, elements that are shown as being directly on, directly connected or directly coupled can be referred to as such. The claims of the application may be amended to recite exemplary relationships described in the specification or shown in the figures.
While certain features of the described implementations have been illustrated as described herein, many modifications, substitutions, changes and equivalents will now occur to those skilled in the art. It is, therefore, to be understood that the appended claims are intended to cover all such modifications and changes as fall within the scope of the implementations. It should be understood that they have been presented by way of example only, not limitation, and various changes in form and details may be made. Any portion of the apparatus and/or methods described herein may be combined in any combination, except mutually exclusive combinations. The implementations described herein can include various combinations and/or sub-combinations of the functions, components and/or features of the different implementations described.
In addition, the logic flows depicted in the figures do not require the particular order shown, or sequential order, to achieve desirable results. In addition, other steps may be provided, or steps may be eliminated, from the described flows, and other components may be added to, or removed from, the described systems. Accordingly, other embodiments are within the scope of the following claims.
Alliez, Pierre, Hemmer, Michael, Portaneri, Cedric
Patent | Priority | Assignee | Title |
11450030, | Sep 24 2019 | Apple Inc. | Three-dimensional mesh compression using a video encoder |
11508094, | Apr 10 2018 | Apple Inc. | Point cloud compression |
11508095, | Apr 10 2018 | Apple Inc. | Hierarchical point cloud compression with smoothing |
11514611, | Nov 22 2017 | Apple Inc. | Point cloud compression with closed-loop color conversion |
11516394, | Mar 28 2019 | Apple Inc. | Multiple layer flexure for supporting a moving image sensor |
11527018, | Sep 18 2017 | Apple Inc. | Point cloud compression |
11533494, | Apr 10 2018 | Apple Inc. | Point cloud compression |
11538196, | Oct 02 2019 | Apple Inc.; Apple Inc | Predictive coding for point cloud compression |
11552651, | Sep 14 2017 | Apple Inc. | Hierarchical point cloud compression |
11562507, | Sep 27 2019 | Apple Inc. | Point cloud compression using video encoding with time consistent patches |
11615557, | Jun 24 2020 | Apple Inc | Point cloud compression using octrees with slicing |
11620768, | Jun 24 2020 | Apple Inc | Point cloud geometry compression using octrees with multiple scan orders |
11625866, | Jan 09 2020 | Apple Inc | Geometry encoding using octrees and predictive trees |
11627314, | Sep 27 2019 | Apple Inc. | Video-based point cloud compression with non-normative smoothing |
11647226, | Jul 12 2018 | Apple Inc. | Bit stream structure for compressed point cloud data |
11663744, | Jul 02 2018 | Apple Inc. | Point cloud compression with adaptive filtering |
11676309, | Sep 18 2017 | Apple Inc | Point cloud compression using masks |
11683525, | Jul 05 2018 | Apple Inc. | Point cloud compression with multi-resolution video encoding |
11711544, | Jul 02 2019 | Apple Inc. | Point cloud compression with supplemental information messages |
11727603, | Apr 10 2018 | Apple Inc. | Adaptive distance based point cloud compression |
11748916, | Oct 02 2018 | Apple Inc. | Occupancy map block-to-patch information compression |
11783508, | Sep 24 2019 | Apple Inc. | Three-dimensional mesh compression using a video encoder |
11798196, | Jan 08 2020 | Apple Inc. | Video-based point cloud compression with predicted patches |
11818401, | Jun 22 2018 | Apple Inc. | Point cloud geometry compression using octrees and binary arithmetic encoding with adaptive look-up tables |
11895307, | Oct 04 2019 | Apple Inc.; Apple Inc | Block-based predictive coding for point cloud compression |
11922665, | Sep 18 2017 | Apple Inc. | Point cloud compression |
11948338, | Mar 29 2021 | Apple Inc. | 3D volumetric content encoding using 2D videos and simplified 3D meshes |
Patent | Priority | Assignee | Title |
10657675, | Mar 06 2018 | GOOGLE LLC | Techniques for improved progressive mesh compression |
6262737, | Jan 30 1998 | University of Southern California | 3D mesh compression and coding |
6879324, | Jul 14 1998 | Microsoft Technology Licensing, LLC | Regional progressive meshes |
6995761, | Jan 14 2000 | California Institute of Technology | Compression of 3D surfaces using progressive geometry |
20060181536, | |||
20130114910, | |||
20190318507, | |||
20190318508, | |||
20190320183, | |||
20200265552, | |||
20200265611, |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
Feb 19 2019 | GOOGLE LLC | (assignment on the face of the patent) | / | |||
Feb 28 2019 | HEMMER, MICHAEL | GOOGLE LLC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 048901 | /0585 | |
Mar 01 2019 | ALLIEZ, PIERRE | GOOGLE LLC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 048901 | /0585 | |
Mar 01 2019 | PORTANERI, CEDRIC | GOOGLE LLC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 048901 | /0585 | |
Feb 08 2023 | INRIA CENTRE AT UNIVERSITÉ CÔTE D AZUR | GOOGLE LLC | CONFIRMATORY ASSIGNMENT | 063026 | /0073 |
Date | Maintenance Fee Events |
Feb 19 2019 | BIG: Entity status set to Undiscounted (note the period is included in the code). |
Date | Maintenance Schedule |
Apr 13 2024 | 4 years fee payment window open |
Oct 13 2024 | 6 months grace period start (w surcharge) |
Apr 13 2025 | patent expiry (for year 4) |
Apr 13 2027 | 2 years to revive unintentionally abandoned end. (for year 4) |
Apr 13 2028 | 8 years fee payment window open |
Oct 13 2028 | 6 months grace period start (w surcharge) |
Apr 13 2029 | patent expiry (for year 8) |
Apr 13 2031 | 2 years to revive unintentionally abandoned end. (for year 8) |
Apr 13 2032 | 12 years fee payment window open |
Oct 13 2032 | 6 months grace period start (w surcharge) |
Apr 13 2033 | patent expiry (for year 12) |
Apr 13 2035 | 2 years to revive unintentionally abandoned end. (for year 12) |